¹ Kuhn, T. S., The Structure of Scientific Revolutions (first edition, 1962; second edition 1970, Chicago: Chicago University Press).Google ScholarDuhem, P., La Theorie physique. Son objet, sa structure (Paris, 1906);Google Scholar English translation The Aim and Structure of Physical Theory (Princeton: Princeton University Press, 1956).Google Scholar

² Kuhn, , The Structure of Scientific Revolutions, pp. 151–152.Google Scholar

³ Duhem, , The Aim and Structure of Physical Theory, p. 187.Google Scholar

⁴ Quine seems to make this slip. This seems to be partly based on a failure consistently to distinguish between ‘Indefinitely many assumptions are needed if any observational statement is to be derived’ (false) and ‘Although “only” finitely many assumptions are needed and so at least one of that finite set must be rejected if the observational consequence proves false, there is no a priori limit on the assumptions that might in turn be affected by that initial rejection’ (true, but unsurprising). It also seems partly based on a flirtation with the idea that not even deductive logic can be taken as fixed here. But if not even a core of logic is taken as constituting the framework of the discussion, then it is not clear that any sense can be made of any assertion about testing.Google Scholar

⁵ An alternative formulation would simply involve an ‘initial condition’ about the total force acting on Neptune at t. But the more complex assumptions about the planets as well as the ‘closure’ assumption would, of course, simply be hidden in such an initial condition.Google Scholar

⁶ For further details see my ‘Falsification, Rationality and the Duhem Problem: Griinbaum vs Bayes’, in Earman, J., Janis, A. I., Massey, G. J. and Rescher, N. (eds), Philosophical Problems of the Internal and External Worlds (Pittsburgh and Konstanz: University of Pittsburgh Press, 1993).CrossRefGoogle Scholar

⁷ The ‘other’ reason given by Popper for the inevitable inconclusiveness of empirical refutations is the alleged inevitable fallibility of basic statements. In fact, as indicated above, this is best treated as another, rather confusing way, of putting the same Duhemian point.Google Scholar

⁸ Watkins, John, Science and Scepticism (Princeton: Princeton University Press, 1984).CrossRefGoogle Scholar

⁹ A quite comprehensive review of both the philosophical and AI literatures on rational heuristic can be found in chapter 2 of Ken Schaffner's recent Discovery and Explanation in Biology and Medicine (Chicago: Chicago University Press, 1993). See also Elie Zahar's Einstein's Revolution: A Study in Heuristic (La Salle: Open Court, 1987).Google Scholar

¹⁰ For the real story of this historical episode see my ‘Fresnel, Poisson and the White Spot: the role of successful predictions in the acceptance of scientific theories’, in Gooding, D., Pinch, T. and Schaffer, S. (eds), The Uses of Experiment (Cambridge University Press, 1989).Google Scholar

¹¹ Popper, BG. This was written before Logik der Forschung but finally published only in 1979. The quotation is from p. 27 and the translation is due to Elie Zahar.Google Scholar

¹² I should not want to assert that scientists are never reduced to ‘random conjecture’ but this is both unusual and very much scraping the heuristic barrel.Google Scholar

¹³ This, incidentally, is why Popper's concession that ‘dogmatism’ mayoccasionally have some value (see ‘NSD’, p. 55 and MF, p. 16) is offbeam. There is never any need for ‘dogmatism’, only a need for good ideas about which particular parts of large theoretical systems need to be amended in view of experimental difficulties.Google Scholar

¹⁴ See his ‘Logic of Discovery or Psychology of Invention?’ British Journal for the Philosophy of Science (1984), pp. 243–261.Google Scholar